Document 11070348

advertisement
LIBRARY
OF THE
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
ALFRED
P.
SLOAN SCHOOL OF MANAGEMENT
THE PRICE EFFECTS OF F.P.C. REGULATION:
COMPARISONS OF REGULATED RESIDENTIAL AND
UNREGULATED INDUSTRIAL PRICES FOR
NATURAL GAS
527 - 71
Paul W. MacAvoy
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
lMBRIDGE, MASSACHUSETTS 0213!
MAY
DEWEY
THE PRICE EFFECTS OF F.P.C. REGULATION:
COMPARISONS OF REGULATED RESIDENTIAL AND
UNREGULATED INDUSTRIAL PRICES FOR
NATURAL GAS
527 - 71
Paul W. MacAvoy
The Brookings Institution, Harvard Law School,
and
Massachusetts Institute of Technology
This essay is part of the author's studies of the
behavior of the Federal Power Coinmission, conducted
at Brookings under a grant from the Ford Foundation.
He is grateful to the Foundation for support, and to
members of industrial economics workshops at the
University of Rochester, Stanford University, and the
ImUniversity of Toronto for counsel and criticism.
portant substantive and procedural assistance have
been provided by David McNicol, graduate student in
the Sloan School and Economics Department of M.I.T.
To this point, at least, he should not be held responsible for what follows.
\3
Lll
RECEIVED
JUN 23
M.
I.
T.
1971
LlbKMrxitS
The interstate natural gas pipelines are in the business
of purchasing natural gas in Texas, Louisiana,
and other south-
west states for sale to industry and municipalities along trans-
mission routes leading to the large population centers in the
The terms of purchase and resale are nu-
Northeast and West.
merous and diverse
—
made more so by futures trading of gas
reserves and by national or state regulation of the sellers and
buyers at each level of the industry
—
but the business is ba-
sically conducted according to the public utility pattern.
pipelines are said to be "natural monopolies."
The
Profits and
prices are constrained by Federal Power Commission regulation
to produce the peculiar American blend of competitive performance
from private companies operating in monopoly markets.
There are theoretical reasons for questioning whether this
is the actual pattern.
The reasons have to do with whether re-
gulation should and could be used to accomplish this goal.
The
pipelines may very well not be "natural monopolies", in the present time frame of reference for pricing, because the operating
costs and capacity-expanding costs of established pipelines appear
1
to be directly proportional to gas throughput.
The state of
1
Cf. S. H. Wellisz, "Regulation of Natural Gas Pipeline Companies: An Economic Analysis", LXXI Journal of Political Economy
33 (February, 1963).
5-^7438
affairs could be such that, not only are the pipelines not "natural" monopolies, but they are not "monopolies".
Historical
patterns of construction of new pipelines in the 1960
dupli-
's
cated earlier lines moving gas to growing centers of population,
so that there were two,
three and sometimes more companies offer-
ing gas to large industrial or retail utility buyers.
There are
conceptual problems in finding the customary public utility "cost
of service" to set regulated prices.
is regulated, because F.P.C.
Only part of the business
jurisdiction is limited to "sales
for resale in interstate commerce"
;
since this legalism excludes
direct pipeline sales to industrial buyers,
joint costs for the
two categories of service have to be divided to find the costs
of service provided by the pipelines to regulated firms.
The
methods for dividing joint costs of installed transmission lines
are unknown and difficult to conceive in the regulatory process.
Indeed, methods could more easily be devised to allow pipelines
to quote prices set by "regulated costs" on "regulated sales"
which would be the same as would occur without regulation.
There are two practical reasons for questioning pricereducing effects from public utility regulation, both based on
present Federal Power Commission procedures.
The F.P.C. sets
profit limits on regulated sales by designating
a
maximum average
rate of return for the seller on the present value of his in-
stalled and working capital equipment; the designated rates of
return have been justified in any number of ways, because they
are the same as those in comparable regulated or unregulated
industries, or are equal to previous rates of this company, or
are half-way between what the company and the F.P.C. staff find
appropriate.
Generally, the maximum allowed rate has not been
restrictive.
In the last reported Commission decision in 1969,
the rate was set higher than the pipeline company has been able
to earn,
for the fourth time in the history of that line; unless
the F.P.C. consistently overestimates both competitive costs and
monopoly profits in this case, the company is
being allowed
under regulation to keep more profits than it has been able to
earn.
This may be an exception, even though it is the latest
decision.
In earlier Commission decisions,
were not essentially any more restrictive:
the maximum rates
the average rates
exceeded estimated marginal costs of capital by more than two
points, even in those years in the late 1960
's
were rising sharply and far exceeded average.
when capital costs
Without regulatory
Cf. P.W. MacAvoy, "The Formal Work-Product of the Federal
Power Commissioners" The Bell Journal of Economics and Manage ment Science Vol. 2, No. 1 (Spring 1971).
,
rules that have the tendency over a decade to reduce average
allowed rates of return to equality with average costs of capital, it can hardly be expected that prices would approach
competitive levels.
doubts.
The second set of findings raises more
Practical problems arise from F.P.C. rules for cost
allocation that are supposed to be the basis for actually setting regulated prices.
These rules, whatever their predicted
effects, have not been applied consistently.
The controversial
Atlantic Seaboard Formula should have imposed costs on unregulated sales
—
whether or not they could be covered by returns
in "competitive" industrial fuel markets
—
that should have
reduced accounting costs on regulated sales and ultimately
regulated prices.
But the Formula has been "tilted" every
which way in case decisions during the last decade, so that by
1968 it no longer applied even on the Atlantic Seaboard Pipeline
Company itself.
2
Inconsistent application of the rules on costs
More precisely, the Formula reduced costs allocated to peak
load service, relative to off-peak service, when compared to costs
that on the
If peak-load service
allocated on a volume basis.
was entirely for rethree days of the year of greatest demand
tail utility buyers, then their costs would be reduced.
—
—
^
404 F 2nd 1268 (1968)
Cf. Atlantic Seaboard versus F.P.C.
and the detailed discussion in A.E. Kahn The Economics of
Regulation (Wiley, 1970) Vol. 1, p. 99.
to set prices, or the bending of the rules to make
ea>
cation an exception, should result in sporadic and weak
from regulation.
If this is the case, the theoretical and
practical problems are overwhelming:
the regulators deal with
non-monopolists in an inconsistent fashion, with no consequent
effects on "costs of service" and ultimately prices.
The means for assessing these possible
—
and contradictory
effects of Federal Power Commission regulation is by comparison
of prices under regulation with those on unregulated industrial
sales.
The next section outlines the logical basis for such a
comparison.
This necessarily involves description and synthesis
of economic models of the effects of regulation,
in particular
those termed "A-J-W models" which were in part derived from
a_
1
priori analysis of regulated pipeline behavior.
is
The description
for the purpose of finding testable propositions about regu-
lated as compared to unregulated prices.
The second section
describes the data to be used for testing regulated against un-
regulated prices, and the third section reports initial tests
regression equations and accompanying observations
—
on the
After H. Averch and L. Johnson, "Behavior of the Firm Under
Regulatory Constraint" American Economic Review 52, (December
1962), pages 1052-69, and S. Wellisz, 0£. cit
.
—
effects of F.P.C. regulation of pipelines.
tentative,
The report, while
leads to the conclusion that the effects are slight
While they follow closely the
but in the expected direction.
directions of the A-J-W theory, the extent of effects from
regulatory constraint is almost unnoticed in the midst of much
more important cost, demand, and institutional factors in pricing.
1.
Hypotheses on Regulated versus Unregulated Prices
The pipeline enterprise faces demands for gas by retail
public utility companies seeking to deliver to home consumers
a
total amount q
for direct consumption q
same period.
and demands
in a year's service commitment,
by industrial consumers during the
Both are contract demands with expressed or im-
plied volumes (implied by commitment of capacity) and the major
considerations are price P
=
f{q-j^)
and P
= f (q
)
•
Conditions
of supply on the regulated sales to home consumers could be mo-
nopolistic or similar to monopoly in price behavior where there
are only two or three pipelines.
Although there are no more
sources of gas for industrial consumers than for retail public
utilities serving home consumers, control of industrial prices
conceivably could be more limited on industrial sales by extensive inter fuel substitution and by the lack of regulatory
mechanisms for reporting and delaying price
"discounts" by any transporter.
curs total costs C = f(q
to provide q
and q
,
q
turns including
The pipeline enterprise infor operating existing capacity
,K)
and for adding to capacity K.
There are
three types of constraints on the pipeline's control of prices:
the first being capacity constraints, the second those implied
by the extent of market competition, and the third in F.P.C.
regulation of profits,
1(a)
Prices without regulation
;
the pipeline enterprise is
the sole source of supply, and maximizes profits subject only
to constraints on capacity utilization.
Gas is difficult or
expensive for the buyer to transport and resell, so that dis-
crimination can be practiced without "leakage" from the low
priced buyer to the high priced buyer.
This pipeline company
has the opportunity to charge different prices for each unit-
volume sold to the same consumer, given that resale can be pre-
vented even on the last units purchased by low-priced users
which might be turned over to high-priced users for the intra-
marginal units.
Different prices can be charged by the expe-
dient of setting different initial lump sum charges for the
initiaton of service which, when added into the marginal prices
and P
P^
R =
,
JP-^dq^^
results in profits
subject to
+ JP2'^'^2 ~
^^^i'^2'^^
q
+ q
^
K.
The company sets outputs and prices according to the first
1
order conditions from
G = JP^dq^ + fP2'^^2 - ^(^1-^2'^^
~
^^^1+^2" ^^
P^ - 0C/2)q
^^^"^
^^^^
- ^ =
P^ - OC/Oq^ -
2^
=
q^ + q2 - K =
-SC/5K
^
+
=
so that relative prices are
P^ = P2 + (S>C/aq^ - dC/Sq^)
(1)
and quantities are
q^
1(b)
+ q^ = K.
(2)
Competitive constraints on pricing
;
these can take many
forms, but the most likely first appearance of alternative sources
of supply would be marked by the disappearance of perfect discrimi-
nation,
a
as the high-priced buyer from one pipeline seeks to become
lower-priced buyer at another pipeline.
The two-part tariffs cannot
This follows closely on W.J. Baumol and A. Klevorick "Input
Choices and Rate of Return Regulation: An Overview of the Discussion" and I. Pressman "A Mathematical Formulation of the
Peak-Load Pricing Problem" both from the Bell Journal of Econo mics and Management Science Vol. 1, No. 2, pages 162-190 and
304-327 respectively (1970)
Here the discussion is centered
on pricing, rather than input factor ratios, for testing purposes,
so that propositions appear to be different while they are not.
,
,
.
be put into effect so that the firm operates as if limited to
G = P^q^ + P^q^ - C(q^,q2,K)
-
q^- K)
l((ci^+
.
The first order conditions are
(1+ 1/e
P
)
P„(l+ 1/e^)
2
- dC/2)q
dC/dq
-
2
-
Zr
=
-
-^
=
2
q^ + q^ - K =
dC/dK +
where
e.
= - (P /q^^) dq /iP
.
.
.
,
=
?5'
the price elasticity of demand,
so
that relative prices are
(1+ 1/e
P
=
1
(1+
—
1
)
1/e
2
)
(dCAq
+
P
(1+ 1/e
-
dC/dq
1
)
(3;
)
2
and quantities are
1
q-
+ q'
= K'.
(4)
These conditions differ if q, =K and q =K in separate periods.
The firm maximizes G= P,q,+ P q^ - C(q, ,q^,K) - 31 (q, -K) - 31 (q„-K)
^
llqz
iz
IX
and the first order conditions are:
p^(i+ 1/e^) - sc/aq^
P2(l+ l/e^)
£.
-
ar^
-
- ^-C/dq^ -
^2
"
q^ - K =
q
-
K =
2
-2)C/^)K +
2'
1
so that relative prices are
(1
+ 1/e
=
P
^
(1 +
1/e
—
ar
=
2
1
)
)
+
(dC/aq
+
P
2
(1+ 1/e
+
1
)
dC/dq
+ dC/aK)
2
.
(5)
10
1 (c)
Prices when only residential sales are regulated
;
the
effects of regulation on pricing of course depend on the
nature of the controls applied by the regulatory commission.
There are many ways of controlling the firm, including setting price directly, but the F.P.C. generally follows policies
of setting dollar limits on profits through the artifice of
finding fair rates of return on investment.
We will avoid the
artifice here in the simple model, by assuming that the Commission holds profits on regulated sales q
- ocC
P q
= M,
where
ot
to M, so that
is the proportion of total costs attri-
buted by the F.P.C. to regulated sales.
The profits on industrial sales are not constrained, so
that with
Here each class of service receives full
(footnote continued)
capacity supply, and costs of capacity are "shared" in the same
way that peak load capacity is shared in optimal electricity
with the sum of marginal revenues equal to long run
pricing
marginal costs. The conditions for setting these prices are
unreal, in that the two classes of sales are segmented by period
and both classes fill the pipeline to capacity; more probably if
not likely, residential sales fill the line in some periods and
k and q^ ^ K,
push industrial sales to other periods. Then
so that without
—
q^
2!f
(1+ 1/62)
=
P
^
1
P
(1+ 1/e^)
-
(ac/dq
+
(1+ 1/e^)
^
dC/bq
)
^
1
dC
(1+ 1/e^)
dK
+
(6)
and the residential consumers pay for all of the additional costs
of capacity.
11
1122
G = P q
+ P q
-
C(q
q z K)
I
-
^^(q
12
+ q - K)
+
the first order conditions are
(1+A)[P^(1+ 1/e^)]
P2(l+ 1/e
)
-
(l+«A)dC/5q^
-
-
(l+aA)sc/dq
-
ZT
^T
=
^1 + q2- K -
P^q^
-(i+o^>)3c/dK +
(1+ 1/e
1
and
=
P
^
(1+A)
with (1+A)
^
,
1
=
5C
/2>C
+
-P
(1+ 1/e^)
(7)
I
2
(1+^)
(1+ i/e^)
^^
^q^
(8)
(7)
,
(3)
for the same market and production
They differ by the factor 1/(1+
and the factor
cost difference
given equation
(1+ot
(^C/dq
(8),
can be
for relative prices under regulation,
compared with equation
of P
(1+ocA)
)
2r
=
1.
Equation
conditions.
—
M
-
- C
-
A)
/ (1+ A)
ac/aq
).
A)
in the coefficient
for the coefficient of the
The first factor is ^
and the second factor is
^
1,
1,
given that
As shown by the Lemma: "if the firm maximizes its total profit
in
subject to the regulatory (inequality) constraint (2)
which s^r,
and if, in addition, the regulatory constraint is
'^
A ^ 1." Baumol
binding and x,^o and Xo^O, then we must have
page 165.
and Klevorick, 0£. cit
(1)
,
,
.
,
.
12
cx-
- 1
(i.e.,
sales).
that not all the costs are attributed to regulated
Thus, regulation results in higher relative prices for
sales coming under Commission jurisdiction.
The higher regulated price, relative to the direct-sale
unregulated price, is not counter-intuitive.
In the short run,
with full use of capacity, the new application of regulatory rules
would require profit reductions on regulated sales that could take
place only by reducing regulated output and consequently increasing
regulated price-
In the long run, the profit constraint applied
to retail utility purchases would, in the A-J-W model, increase
the profitability of capital relative to other inputs and thus the
level of total investment.
If there were economies of scale,
in-
creased capacity would reduce marginal costs and thus the absolute
level of prices on both regulated and unregulated sales.
The in-
creased capacity would be more fully devoted to the unregulated
sales because, whatever the relative levels before the application
of regulation, marginal output in the unregulated sector would now
be relatively more profitable.
being lower [as shown by the
This follows from attributed costs
(l+otX)
term on unregulated costs] and
from relative profit reduction on regulated sales
(as
shown by M)
Marginal output in the regulated sector should have reduced profits,
in other words,
so that there should be a long run relative decline
in regulated sales and a relative increase in regulated prices.
13
1 (d)
Prices when regulation requires specific cost allocation
procedures
;
the rules for apportioning costs between regulated
and unregulated sales are more complex than suggested by
the
(1+aX)
oC
in
expression, and the added complexity may directly
affect quantities and prices.
This is likely with respect to
the regulatory rule applied with wide discretion by the F.P.C.,
the Atlantic Seaboard Formula for pipeline cost allocation.
costs on sales q
Only
going to local public utilities are used to
find regulated "costs of service" and prices.
The designated
costs for delivery on these sales are made up of a portion of
total variable costs C
(v)
,
which Atlantic Seaboard prorates on
the basis of relative volume q /(q-|+q2)/ and a portion of capital
costs C
(K)
which the Formula prorates to regulated sales only
partly on the basis of volume.
The capital costs are divided
into two equal parts and prorated as follows:
half is attributed to regulated sales q
so that regulated capital costs are C
(1)
the first
according to volume,
(K) -q
/2
(q-j^+q
)
;
(2)the
second half is attributed to regulated sales according to peakload volume so that, if these sales q =
_/3K
at peak capacity K,
then regulated capital costs also include C{K)j3/2.
The Atlantic
Seaboard Formula limits earnings to
P^q^ - C(v) •q;L(^l+^2) " C
(K)
q^/2 (q^+q2) - C (K)//2
^ M.
14
When this regulatory constraint is applied to the profit function,
it should have some effect on relative price P
profits R = P q
+ P ^(1^ ~ C(q-, ,q^,K)
Consider
.
subject to the Atlantic
Seaboard Formula and to the capacity constraint
- K.
q, +q„
Then the first order conditions are
(1+X)P
- dC/dq^
(1+ l/e,)
-L
1
2
-
M-(^C(vydqJq /(q +q
1
/{q+a)
C(v)q
P^{1+ l/e^)
+
1
•-
-
1
C(K)q /2(q +q
2
2
1
12
2
2
-
=0
?C
bC/dq^ + A[-(c)CCvydq2)q^/(q^+q2) +
C(v)q^/(q^+q^)
P^q^
-
]
)
-
)
2
C(v)q /{q +q^)
- C (K)
+ C (K) q^/2 (q^+q^)
+ q
1
-
=
zT
q^/2 (q^+q^) - C (K)y3/2
q
-bC/dK
-
]
=
-
M
K
=0
^
=0
2
Mq^+/3q^+^q2)/2(q^+q2)-^
+
so that relative prices are
(1+ 1/e
^
1
)
=
P
+
P
- dC/dq^)
ibC/dq
(1+A) (1+ 1/e^)
2
(1+X) (1+ 1/e^)
'"SC(v)
^
C(K)/2(q^+q^)
L
aq^
(9)
ac(v)
[q^/(q3^+q2)]
(1+X) (1+ 1/e^)
+
2
dq^
,
+ C(v)/(q^+q^)
+
15
a rather special version of the general price results from regu-
lating only one class of service.
Here the regulated price is
less than that from allocation of costs according to the
posited for equations
(7)
and
(8)
cL
rule
under certain conditions.
If
the second and third terms on the right hand side of equation
(9)
together are less than the second term on the right side of equa-
tion
oc
^
(7),
this will be the case.
"^i/Cq-i+q
)
This is to require that, assuming
and the \ constraint lies between
the average variable costs C(v)/(q +q
costs C(K)/2(q +q
should hold.
)
)
1
and
0,
then
along with average capital
both have to be positive.
These conditions
The sense of this is that forcing some of the costs
of additional regulated output over onto the unregulated output
has the effect of reducing the regulated price.
1(e)
Rate-of-return regulation
with an arbitrary
oC
;
merely setting rates of return
is less ambiguous in its effects on relative
prices than regulation by Atlantic Seaboard cost allocation.
This
procedure has the effect of raising the regulated price at least
when the regulated sales are at a farther distance from gas field
reserves.
P q
and X
Consider the profit function to be
+ P2'3o - ^1^1 ~ ^2^9' ^^^^^ ^1 ^^^ capital resources
non-capital resources available to the regulated firm at
prices r^, r2 respectively.
'"
The restraint on profits is that
This follows directly from Baumol and Klevorick, 0£.
cit, .
16
P q
- o'-s^x
- oir X
= M,
where
return on capital such that
s
s
- r
is the allowed rate of
= V - 0.
The first order
conditions for
G = P^q,- ^2^2"
^l^l" ^2^2~
2^(qi+q2-^l)+
H^ ;^q^-<^s^x^-oir ^x^-
M)
include
(1+X)P^(1+ l/e3^)-(l+o^X) (r^dK/aq^+ r2aL/dq^)
- (l+<x>)
(s-r^)aK/dq^ =
P2{1+ l/e^)-{l+ocXj {r^i>K/dq^+ r2aL/d q^) - (1+«A) (s-r^)
If marginal costs dC/Sq
1
1
(1+X)
(1+X)
(1+ 1/e
(1+
)
•V
l/e^) ^
(1+ 1/e^)
are defined as
(r
SK/dq + r dh/bq
dK/feq
),
2= 0.
then
(l+cxTg
(10)
(l+X)
^^
(1+ 1/e^)
""l^'l^dq^ '
sq^
dq^J
aq2
the same expression for relative prices as in
(8)
,
but with the
addition of the last term on the right hand side of this equation,
This last term is greater than zero if aK/2>q
—
^ BK/dq
1
if the
2
distance from purchase of the gas to final delivery is greater
for regulated sales than unregulated sales,
for one
—
so that
the relative price of regulated gas transmission is even greater
when rate of return limits are set only on the regulated sales.
This last term is negative if dK/Sq
^
SK/dq
,
or if the unreg-
ulated sales occur at greater distance from field reserves, so
17
that relative regulated prices are reduced even more from
this type of regulation.
The pattern of residential-indus-
trial location sets these relative capital expenditures; but
given that they have taken place, then their effects on relative regulated prices are always greater under this regimen.
These are equilibrium conditions for relative prices,
assuming that demands and costs are known and that regulation
is continuous as well as certain.
These assumptions do not
always hold in all markets for gas transmission, and do not
hold often enough that questions can be raised on the usefulness of an analytical apparatus based on them.
The answers to the questions are not well developed at
this time.
They depend on the extent to which models with
uncertainty and with dynamic regulatory conditions differ from
the certainty-static models, and these more complex models re-
main to be developed.
At this point, of the many promising
but only partially developed analyses, those based on condi-
tions most similar to actual market and regulatory conditions
differ little from the certainty-static models.
for this assertion are two.
The reasons
First, with respect to uncertainty,
cost and demand conditions different from those expected must
occur at all levels of prices and volumes of service, but it
18
seems plausible to assume that variation from expected cost
and price levels is going to be greater when gas demand is
highly elastic with respect to prices of other sources of
energy.
These demand conditions should occur most widely
—
in unregulated markets
almost as
a
matter of classifying
these markets as competitive enough not to be regulated.
That is, uncertainty has the greatest effects in unregulated
markets.
Second, with respect to regulation,
the commissions
operate with a lag, but not an extended lag, given that F.P.C.
staff seeks to bring about price adjustments resulting
profits at the regulated level within the year
.
in
That is, the
imperfections of regulation may not require variations in the
model structure for year-to-year analysis.
Then ex post pro-
fits should come close to ex ante regulated profits,
categories of service
—
the regulated categories
—
on the
least
subject to uncertainty.
At this point, the rather extended array of price varia-
tions might well be summarized.
The price hypotheses differ
quite markedly from each other, when there are differences in
degrees of market control and in the nature of regulation.
These differences are shown in Table
1.
When it is possible
19
20
for pipelines to charge a different price to each retail dis-
tributor or industrial buyer, and there is no regulation, elasticities of demand do not enter the coefficients of the equation
P
= aP
+ b (DC/aq
the first row.
-
When,
dC/dq
)
,
as can be seen from reading across
for reasons of rivalry between pipelines
in the same market, prices turn out to be uniform to all indus-
trial or all retail utility buyers
all buyers of q
—
—
to all buyers of q
or
the relative elasticities of demand enter
both coefficients a and
b,
as shown in the second row.
The price
behavior of perfect competitors is shown in the third row, for
contrast.
Here again, there should be contrast, because in the
long run, profit constraints, if effective, can only cause firms
to leave the industry.
Regulation results in testable differences in coefficients
the last three columns in Table
regulatory constraint
X
1
show this.
These
A.
Terms including the
are found in both a and b, and additional
variables also enter the equations with
and d.
—
X
terms in coefficients c
terms increase the size of the coefficients, ex-
cept in "cost-allocation regulation," so that the regulated price
is increased relative to the unregulated price,
and transport cost
differences are "marked up" with a regulatory premium.
.
21
Data for Testing Regulated Against Unregulated Prices
2.
The information on prices charged by the interstate pipelines is voluminous, complicated and not altogether useful for
testing the effects of regulation.
The regulatory process may
itself be responsible, since the demands of the F.P.C. for information are for detailed statements on assets and liabilities
at specified
(at
least annual) time periods, but for less de-
tailed statements on flows of expenditures and income by contract, or on amounts of gas delivered.
Also, the demands are
for rather complete information on regulated but only partial
information on unregulated sales (for cost allocation purposes)
There are data on volumes of gas delivered under contracts with
the regulated retail gas distributing companies, and in the
Form
2
Annual Reports to the F.P.C. on the "demand charges"
(an-
nual or monthly charges for providing service, where these charges
are not related to the rate of delivery at that time) and on the
"commodity charges"
retail gas sale.
(the prices per mcf of gas delivered)
There are annual statistics in the Form
on each
2
Reports
on the volume and total revenues from unregulated direct industrial
sales; these have been supplemented by interviews and special
surveys of particular pipelines, to find the actual prices charged
22
on the average rate of delivery for unregulated sales in 1969.
The price information is to be used to test the coefficients a,b in
P-|^=
aP2 + b (3C/3q
- 'SC/Sq
)
for the effects of
pipeline and inter-fuel competition, and for the effects of
regulation.
Since this expression has been derived with respect
to marginal prices, the commodity charge on regulated sales is
designated to serve as P
ted sales should be P
.
and the commodity charge on unregula-
Without a commodity charge on direct
sales, however, the average or unit price is used and designated
P
.
With P_ = P
+ w/q
where w is any additive to the marginal
(discounted) charge on unregulated sales, the equation is still
the same
—
that is, P
as given above
—
= aP
+ b (3C/aq
-
QC/dq
)
+ d with a,b
but with the possible addition of a constant
1
term d.
The tests call for computation of a and b by least squares
regression, so that data are required on the marginal costs of
transport
Z)C/dq,
and 9C/aq
as well.
The tests require independent
This equation can be derived by substituting ^o ~ ^2 *" ^/^ ^^
the appropriate objective functions in the preceding section of
this paper.
If Sw/Sq^ ^0, or charges in fact do vary with the
amount of gas delivered, then an additional term appears (for
example) in equation (3) above.
23
estimates of e
e.
and e
so as to be able to separate the X and
effects found in both a and b.
These requirements will be
discussed and met in turn.
2 (a)
The marginal costs of transport ;
these costs can take on
many different values, and the proper measure depends on the re-
levant time period, and location of the pipeline, and the unit
prices of capital and of pumping fuel at the time estimates are
made.
For purposes of studying recent prices, the time period
chosen is a "recent year" in the interval 1952-1967 and an ob-
servation is "expenditures on operations, maintenance, administration, transmission, storage and depreciation" in one of those
1
years for each of eight large interstate pipelines.
The pipe-
lines have been chosen for similarity in the timing and techno-
logy of construction of the main pipeline; each had more than
1,000 miles of total pipeline,
75 per cent of which had diameter
equal to or greater than 15 inches, and the main lines were con2
structed over roughly the same types of terrain.
The length of
The Federal Power Commission Statistics for Interstate Natural
Gas Pipeline Companies 1952 through 1967, annual volumes.
2
The pipelines were: The Atlantic Seaboard Corporation, Colorado
Interstate Natural Gas Pipeline, Michigan-Wisconsin Pipeline Company, The Natural Gas Pipeline Company of America, Tennessee Gas
Transmission Company (a division of Tenneco) Texas Eastern Transmission Company, Transcontinental Gas Pipeline Corporation, and
Trunkline Gas Corporation.
,
24
the main line of each transporter was estimated by inspection
of maps and engineering drawings, to provide data for the var-
iable "m = mileage", and throughput of gas was recorded as "q =
1
volume."
The most general equation form chosen was C = ol+ Am-q,
which when fitted to the sample, proved to be C = 44.7(10
R
2
=
.
958
with
ex
and
j5
+
)
.57m.q,
in cents per thousand cubic feet of gas
transported for 100 miles.
Attempts were made to account for
differences in factor prices among the transporters, and for the
"mix" of combinations of volume and mileage, as well, but they did
not improve the fit as measured by R
The introduction of var-
.
iables to account for the year of the observation and/or to ac-
count for the particular pipeline in that observation did not
change the values of
oC
or
ji
nor increase the value of R
2
The
.
measure of marginal costs is .57 cents per hundred miles, so
that values of transport cost differences
taken to be .57(m -m
)
where m
and m
(oc/aq
- z>C/2)q
)
are
are mainline mileages
from a common gas field origin point to the points of resale of
regulated gas
qj_
and unregulated gas q
.
Volume is recorded as annual sales in F.P.C. Statistics
,
op
.
.
25
The elasticities of market demand ;
2(b)
the price-sensitivity
of gas demands of home and industrial consumers has been dis-
cussed and analyzed in a number of recent studies, but mostly
for the purpose of analyzing the effects of regulatory policies
in the field producing areas.
There have been no studies of
the specific elasticities of demand of those buying gas directly
from natural gas pipelines.
In order to find first crude esti-
mates of e,,e_ here, a sample of contract prices and quantities
has been constructed and a regression analysis of the sample
completed with equation forms suggested by the more detailed
2
studies
The markets for gas in which the pipelines are sources of
supply probably encompass whole metropolitan regions in most
cases.
The purchasing retail distributors can go to alternative
sources of gas that deliver within their metropolitan regions.
Cf. J. D. Khazzoom "The F.P.C. Staff's Econometric Model of
Gas Supply" Bell Journal of Economics and Management Science Vol.
2, No. 1 (Spring 1971) and P. MacAvoy "The Regulation-Induced
Shortage of Natural Gas" Journal of Law and Economics (Spring 1971)
,
2
See P. Balestra, The Demand for Natural Gas in the United States
(North Holland, 1967) and H.S. Houthakker and L.D. Taylor, Consumer
Demand in the United States 1929-1970 (Harvard University Press,
1966)
.
26
whether pipeline company or LNG company; the industrial consumer
can go to the same sources, as well as suppliers able to provide
large volumes of coal and fuel oil,
in the same locations or with
access to transportation to these locations.
Data on 1969 prices
of all fuels, and on the contract volumes of delivered natural
gas,
have been collected for 17 metropolitan areas in which there
were purchases by retail public utilities, and for 49 locations
in which there were purchases by industrial buyers.
With a single year's data, the demand function takes the
form
log(q^)
=
ex
+ /31og(P
.)
+ -^log (P^)
+ Slog (Y)
consumption of consumers
where q.
^
^1 is annual qas
ginal gas price to these consumers, P
fuel,
F
i,
+ e.log(N)
P
gi
+
is the mar-
the price of an alternative
and Y and N the "market size" variables such as per-capita
income and population.
The residential demand function, as found
by fitting a least squares regression, proved to be:
log(q.)
^
= 8.106 - 1.907 log (P
)
"^l
(1.342)
+ 0.390 log (P
(0.816)
)
-
0.445 log (TTD)
0.545 log
(Y)
(2.792)
2
2
+ 0.564 log(N)
(0.337)
+
^
;
R
=
.374
(11)
(0.711)
That is, in the absence of more than one year's data, lagged
terms in q, P to account for adjustment processes had to be deleted.
the
TDD denotes "temperature degree days" in 1969 at each of
17 locations.
27
a
relationship noted for its general agreement with empirical
—
regularities in economics
gas price decreases have the ef-
fect of increasing quantity demanded,
and income, population
and fuel price increases have the effect of increasing the annual quantity demanded.
But the regularities are not strong:
the coefficients are not statistically significant and the e-
quation only explains 37 per cent of the variation in the dependent variable.
quite high
range
—
—
The price elasticity, in particular, seems
the coefficient is -1.907, clearly in the elastic
but it is not statistically significantly different
from zero.
The estimate is subject to change; after collecting
1958 data on q
and N for these 17 locations, the regression
was refitted for q
11
= f[q
ft-1), P ,,
that the coefficient for log
gl
(P
)
N.
t-i
^
,
.
.
.
.)
with the result
equaled -0.950 and for log q-,(t-l)
2
equaled +0.970; both were statistically significant, and R = .998.
The conclusion must be that the residential distributor elasticity e
lies between zero and -1.9, with the most likely value
between -1.0 and -1.9, so that demand is "elastic".
The 1969 data on industrial demand proved to be much more
extensive.
Price and quantity information have been collected
on 49 direct sales, and using corporate employees as a measure
28
of "size of market," the regression equation was found to be:
log{q„)
^
= 11.372 - 1.785 log (P
)
5"^
(0.980)
- 0.249 log(S32)
-
(12)
(0.114)
0.144 log(S33) - 0.205 log(S29) + 0.681 log(EES);
(0.168)
(0.125)
(0.192)
2
R
= 0.306
Here the coefficients of P
cant,
„
g2
and EES are statistically signifi-
and the first of the dummy slope coefficients
(for industry
classification 32) is also different from zero. The remaining coefficients do not differ from zero, and R
2
indicates that only
is explained by the regression.
30 per cent of the variance in q
Demand is somewhat more elastic than for retail distribution:
the elasticity for the base industry (chemicals)
industry 32 (stone-clay-glass) is -1.785
industry 33
(primary metals)
is -1.785 -
-
is -1.785,
.249 = -2.034,
.144 = -1.929,
industry 29 (petroleum refining) the estimate is -1.785
-1.990.
for
for
and for
-
.205 =
Each of these is different from zero, and seems to be
close to -2.0, approximately twice the elasticity of retail dis-
tributors demand.
Further regression analysis suggests that industrial demand
is more elastic
tutes.
—
particularly where there are close gas substi-
Data were available for
a
more limited sample of 39 gas
.
29
price and quantity observations (excluding sales to companies
in the petroleum refining industry) that included industrial
coal prices at those locations, and the regression from this
sample was:
log(q
)
= 10.954 - 3.471 log(P 3)
- 0.962 log(S32)
^
(0.918)
1.007 log(S33) + 1.578 log
(0.673)
-
(13)
(0.542)
(P
)
;
R^ = .378
(1.346)
Here the estimated elasticities varied from -3.47 (on sales to
firms in the chemicals industry) to -4.54
in the primary metals industry)
.
companies
(on sales to
Gas demand is less elastic
where the use of gas is as a process raw material with poor substitutes
(as in chemicals)
and more elastic where the use is as
a boiler fuel with good substitutes
(as in metals)
Equations have also been fitted for two of the industries
separately (given that there were sufficient data)
.
These show
substantial price elasticities, and also show differences between
them sufficient to explain in part the low values of R
2
in the
A more complete but not necessarily illuminating form is as
follows:
iog(q /EES) = 0.757 - 2.168 log (P*2/Pf) " 1.109 log(S32)
^
(1.175)
- 1.049 log(S33)
(0.689)
where P^ equals price of coal and P
^
(0.542)
+ 2.958 log (P /P
(0.821)
)
;
R^= -364
equals a fuel price index.
30
combined-industries equations.
The equation for demand for gas
in the chemicals industry is
log(q
)
^
= -2.021 - 3.486 log (P* )
5^
(2.118)
4.049 log(Pc)
+ 0.839 log(EES)
+
(14)
(0.315)
R^ = -567
;
(1.938)
for 12 observations, and demand in the stone and glass industries is
log(q_) = 11.641 - 2.778 log (P*
2
^
(1.371)
1.163 log(P^)
)
+
(15)
(0.239)
R
;
+ 0.819 log(EES)
= -232
(0.998)
for 19 observations.
They indicate values of e
= -3.49 and
-2.78 respectively, and as such add to the impression that in-
dustrial demand elasticity lies in the range from -2.0 to -3.5.
This range, along with values of
e
be used in the calculations for the
2 (c)
between -1.0 and -2.0, will
X.
effects from regulation.
Elasticities of demand and individual pricing decisions
;
the elasticities of market demand may not determine individual
pipeline pricing decisions, when there are a number of pipelines
serving the same metropolitan region.
These differ from market
elasticity in some circumstances for any one transporter because
prices reduced below the joint-profit-maximizing level result in
a
disproportionate market share which can be maintained for
a
31
long enough period to more than compensate for the reduction.
The greater elasticity could provide the incentive for lower
prices on either residential or industrial deliveries, so that
final contract prices are below those consistent with e
,
e
.
How great the further reduction should be depends on the oli-
gopoly firms' interaction, and there are any number of theories
on the results of that interaction.
These,
if testable,
show
relations between relative firm size, the quality of information
on prices, and approach a lower bound on price designated by the
This replaces
Cournot theory.
e.
by ne
.
for "n" equivalent sized
firms in all cases described above, where now ne
.
is equilibrium
market elasticity for market-wide price determination (even though
1
individual firm elasticity is larger at each stage of price setting)
A first measure of the extent of any such interaction,
of replacements for e-^,e^ as ne
or ne„ where n -
in terms
in the rela-
1
tive price equations, can be made by observing the coefficients
in price equations where no regulatory effect enters.
Consider the relative price equation P
*
where P_
<ia
Cf
.
*
and P
2b
W. Vickrey,
= aP^,
+ b (BC/bq2a~^'-^2D
are both samples of industrial prices,
Microstatistics (New York, 1964)
,
one for
pages 337-38.
32
industry a and the other for industry b, and 2>C/aq
2a
,
'&C/2)q
2b
are the marginal costs of transporting to each of the points in
the sample.
There are 49 observations of 1969 annual sales of
four pipelines to companies in the chemicals industry and the
electricity generating industry.
icals and P2 (49)
are ^C {28)
/l>q^^
for electricity generating,
and 0C(49)/3q
P2(28)
for chem-
(28)
the marginal costs
for transporting along any one of
With the relationship fitted in the regres-
the four pipelines.
SI on form
The prices are P
1
-
P*(49)
=
AP2= (a-l)P2(49)
+ b [aC (28) /2>q2- &C(49)/dq ]+ d
the first comparative equation for price is
AP = -0.633
^
P (49)+ 2.231[dC(28)/dq
2
(0.086) 2
(0.662)
R^ = .608
-dC(49)/dq
]
+ 276.7;
(16)
2
and the implied elasticity of demand of buyers in chemicals from
a
single pipeline is b = l/(l+ei) or
e-,
= -1.83.
The second com-
parative price equation is [for P„(29), petroleum refining, against
P
(32)
stone-clay-glass]
AP = -0.957 P^(32) + 2.071[&C(29)/aq
2
(0.026)
R^ = .993
(0.149)
-
ac(32)/6q^j + 273.755
"^
(17)
1
The constant term accounts for pipeline-to-pipeline differences
in transport costs, providing a rough net difference due to depreciation expenses in the different transporters.
:
33
and the elasticity of demand b = 1/(1+62)
^°^ ®9 ^" petroleum
equals -1.93.
The third comparative price equation is as follows,
for
39 observations of contracts with stone-clay-glass manufacturers
(32)
and the electricity generating companies
AP^ = -0.784 P*(49) + 2.303[«C(32)/aq
2
(0.175)
R
^
(0.881)
(49)
_
^
-ac
(49) /©q^] +
2
305.638
= .359
(18)
with the elasticity of demand equal to b = 1/(1+ 1/e
All three equations suggest that n ^
1,
)
= -1.76.
or that the elasticity
of demand for the purposes of individual pipeline price setting
is the same as the industry-wide demand elasticity.
There would
seem to be no price-reducing effect from the presence of a second
(and perhaps third)
pipeline on the supply side of most individual
gas purchase markets.
The price setting practices in retail utility gas markets
seem quite different from those in industrial markets.
The pipe-
lines have long set two-part prices on contracts with retail utility
companies, under the jurisdiction of the Federal Power Commission.
The marginal price or "commodity charge" has not been set at a level
in keeping with the residential market elasticities of demand, but
as if marginal prices were being used to add to marginal consumption
34
while the initial or demand charges determined the bulk of the
profits.
This can be seen from using a sample of 26 paired resi-
^P= b (OC/aq
dential sales for 1969 to fit the regression
for sales at locations a and b.
2
R
=
First, AP = +0.869
.312 and given that b = 1/(1+ 1/e^)
of demand e
is undefined.
,
-
ac/&Q
la
(3
C/aq
^]b
-
dC/Zq
the implied elasticity
This would seem to suggest that prices
are in the inelastic range of the demand function.
However, this
would not seem to be the case for the combined initial and marginal
prices together.
To the contrary,
a
second regression was fitted
2
as follows:
AAR
=
+2.4510C/dq
- aC/S9q );
R
= .317 for AR equal
to the average revenue from demand and commodity charges, and the
elasticity implied by the fitted value of b is -1.70.
The two re-
gressions indicate commodity prices in the inelastic range of the
market demand function, but the average of demand and commodity
prices comes close to the equivalent "best profit" single charge
if market demand elasticity were close to -1.70.
This is not un-
likely; the value of -1.70 is within the range of elasticities
observed for retail gas utility markets in 1969.
Here P1-P2 = AP = (a-l)P2 + b(&C/Qqi - ZC/Qq,2) + d, with a=l
d=0 because the samples ^-,,q_2 both have the same price elasticity and are carried on the same (or almost identical) pipelines
.
)
);
35
The contrast between industrial and retail gas utility
pricing by the pipelines is fairly evident, even if based only
on first round evidence.
set with regard to market
ticities.
The prices for industries seem to be
(rather than individual) demand elas-
This is to suggest that the number of pipelines at
any industrial consumption location does not make a difference
for the level of prices there.
for retail utilities,
Two-part tariffs seem to be set
so that the marginal retailer's price
more closely approaches marginal costs than do industrial prices
approximate marginal costs on those sales.
the analysis of regulation,
(1)
This is important for
since it leads to two conditions:
the industrial elasticities contained in a,b in the relative
price equations are within the range -2.0 to -3.5;
(2)
the re-
tail gas utility market elasticities should be between -1.0 and
-2.0, but two-part pricing makes these elasticities for indivi-
dual pipeline marginal pricing effectively equal to zero.
3.
The Testing of Regulated Against Unregulated Prices
The expectation is that profit or rate-of-return regulation
increases the prices of regulated sales relative to unregulated
sales.
The more stringent the regulation
—
the closer gross
36
revenues come to costs, or the closer the allowed rate of
return on capital comes to the marginal costs of capital
—
1
the more apparent the relative price difference should be.
The place to look for these effects is in the coefficients a,b
in P^ = aP^ + h{^C/^q_
f
[
(1+a^)/ (1+X)
]
- ac/Sq^)
,
since a = f[l/(l+>.)], b =
and X ranges from zero without regulation to
-1 with the rate of return set exactly at the marginal cost of
capital.
Now that other factors in a,b have been assessed and
-
presumably quantified, this approach to finding the regulated
price effects can be undertaken.
The full sample of contracts of five pipelines with in-
dustrial and retail utility buyers can be used to fit
a
regres-
sion of regulated marginal price P, on unregulated unit price ^2'
the difference in costs of transport
constant.
P
- '9C/8q
(dC/dq,
)
,
and a
In this least squares equation
= aP2 + b(ac/dq^ - ac/dq^)
+
d-LL^^
+ d2L2 + d3L3 + d^L^
The constant term is a series of dummy variables,
four of the five pipelines
L-j^
(to account for the effects of
Cf. W. Baumol and A.K. Klevorick,
0£. cit
.
L
for
37
differing field gas purchase prices on the absolute levels of
resale regulated prices)
The observations have been paired
.
randomly to make up a sample of 137 and the least squares re-
gression from this sample is:
P
2
= .730P* + 1.296(ac/aq
-
18.703L^ + 84.632L
3
-
©C/dq^)
+ 8.923L^ - 62.551L
(19)
.
4
The price relationship would seem to show lower prices on regulated sales, and higher prices on sales with greater transport
costs.
There are indications that this is a superficial view
of the pattern, however.
—
—\— 4—^
if a =
(l+>0 [1+ 1/e-^l
given above), then
A
The coefficient a is perhaps too large:
and e„ = -2.0,
e^
2
= -.31,
1
undefined (for reasons
and F.P.C. rate of return controls
have succeeded in somewhat increasing relative prices on regulated
The logic of this can be seen from separating pipeline costs
into (a) field purchase price, and (b) transmission costs, and
then deriving first order conditions for maximum profit resale
prices. The field purchase price enters as a constant in the
expression for P
This constant will differ for each pipeline.
2
This regression was fitted with P
as follows:
*
*
P = 1.372P - 1.778(dC/aq-,^
(0.060)^ (0.360)
as the dependent variable,
iC/dq^)- 12.324L + 85.725L
(19.560)-^
(17.587)^
25.669L - 116.116L.;
R = .109
(19.493)
(27.745)
The reason for the reversal of variables was that more serious
errors in data were expected to occur on unregulated sales, and
these could better be relegated to the dependent variable.
+
38
On the other hand,
sales.
if e
= -3.5 and e
= 0,
and the effects from regulation are negligible.
then
X=
-.02
Within the range
of "reasonable market demand conditions", the residual effects
from regulation are
negligibly to moderately price increasing.
The process of regulatory control here might be viewed as
taking place in two steps.
First, the price setting procedure
under regulation iir/olves two-part tariffs, which increase initial
charges from zero to some higher amount while reducing marginal
charges.
The reduced marginal prices appear to be lower than
comparable unit prices on unregulated sales.
Second, the rate-
of-return review procedure raises the marginal regulated price
above the level that that marginal price would have without
regulation but with two-part tariffs.
The second step alone
can be said to involve a relative regulated price increase, re-
ducing regulated sales.
The other parts of the equation are revealing of the price-
setting process, and perhaps of the effects of profit regulation,
as well.
The coefficient b for (ac/2>q
quite large
—
- ^C/bq,2^
appears to be
there seems to be much more price variation with
distance than called for by "costs of transport" differences.
This could be entirely
a
matter of appearance, caused by under-
39
estimation of ^C/'bq
and ac/^q
that the value of
the portion of costs allocated to regulated
sales,
with
e,
01,
is very high,
=
and
.
since b =
[
X estimated from
with b are close to
.8.
Another explanation could be
(l+ot>)
a,
/ (1+^)
]
•
[1/(1+ i/e
)]
and,
the values of oC in keeping
This is to suggest that 80 percent of
costs are allocated to regulated sales
—
a percentage that is
not inconceivable, but somewhat removed from present practice.
The third explanation comes from closer examination of the data:
it appears that sales at the "farther distance" are predominantly
sales to retail public utilities, and the relatively higher prices
on these sales, resulting from regulation, are confounded with
This explanation of
the "distance" effect.
b,
which seems most
likely, is an indirect test of "Atlantic Seaboard" and "rate of
return" effects from regulation,
1
effects are confounded with b.
since coefficients to test these
If "Atlantic Seaboard" prevailed
These variables, P
&C/dq. L, confound attempts to measure
the effects of more complicated regulatory procedures, such as
"Atlantic Seaboard" cost allocation or "direct rate-of-return"
(as in equations (9) and (10)
Further work will have to be
done to deal with complicated cost allocation and rate of return
regulation directly, because the relevant variables when specified turn out to be collinear with the transport cost variables
used here to find b. At this point, there is no evidence of
leveling of prices with respect to costs, as expected from "Atlantic Seaboard" allocation; indeed, the last paragraph was to
the contrary.
It is not likely that as the case decisions indicate, the Formula is not consistently honored in the costing
and pricing decisions of the large pipelines.
,
)
.
,
40
the coefficient of b should be "too low", but if "rate of return"
regulation with arbitrary
ot
allocation prevailed, then the b co-
efficient should be "too high".
In fact,
it is too high,
so that
none of the regulatory constraint effects from regulation are
shown.
There are a few further tests possible with the remaining
The different pipelines have entirely different
coefficients.
price levels for home sales, perhaps because of different degrees of regulation applied by the F.P.C. but more likely because of historical chance in the purchase of field supplies of
gas at lower or higher prices than paid by other pipelines.
The effects on retail consumers versus particular regulated
industries are similar to those shown by the pooled industryBut the differences all tend to show less ef-
retail sample.
fects from regulation.
For unregulated prices to buyers in the
chemicals and stone-clay-glass industries, versus regulated prices,
the regression equations were as follows:
Chemicals,
P* = 1.576P
2
1.714(ac/dq,
-
(0.065)-^
N = 38
-
dC/dq
2
-
7.476L^;
(23.777)-^
)
^
(0.454)
R
= -687
-
45.493L, + 50.294L
(18.387)^
(20.440)
(20)
41
Stone-Clay-Glass, N = 35
= 1.459P
P
2
1.828(dC/aq
-
(0.082)1
-
3C/3q„)
*
(0.513)
46.388L + 56.086 L
^
(26.664)1
(24.545)
(21)
= -644
R
+ 1.382L^;
(31.383)''
-
which imply values of X somewhat lower than those derived from the
all industries-all retail regression:
and X
~
electricity and metals industries versus retai
similar to that from the pooled data.
Electricity
= 1.015P
P
2
1
(0.547)
Primary Metals,
1
utility buyers are
They are as follows:
^
)
+ 13.865L
(30.034)1
+ 149.650L
(22)
(24.948)^
N = 37
P* = 0.858P - 0.432 (ac/2>q
^
(0.045)^
(0.886)
2
(29.999)
= -2.0,
R^ = .841
+ 31.237L •
(25.368)-^
+ 215.448L
-.2 for e
N = 27
0.493 (2>C/aq, - 'BC/aq
-
(0.092)1
~
The equations for primary buyers in the
= -3.5.
for e
here X
;
R
-
dC/Bq
= -.307
)
+ 147.609L
+ 215.586L
(27.985)-^
(31.974)
.
-^
These two regressions show slight effects from regulation, as
slight as the all-industries regressions, even though the values
of the a coefficients are lower.
This occurs because the cal-
culated elasticities in these industries are less
(e
^ -1)
.
(23)
42
imply values of
The coefficients for P
"
\
close to -.30, and
1
any great differences would be difficult to take seriously.
The basic question
—
whether regulation raises relative
prices, or whether the effects of
X
constraints can be observed
-
cannot be dealt with unless the influence of the two-part tariff
is separated from the effects of rate-of-return controls.
This
is necessary because there may be a number of reasons for two-
part tariffs on sales to retail utilities.
They are in keeping
with state regulatory requirements on the buyers' resale prices.
It is conceivable that they serve as an institution to provide
"peak-load" insurance at the cheapest rate.
To make the two-
part tariff an effect from F.P.C. regulation alone confuses the
assessment of that Commission's activities.
There is no reason for regulation to appear to be more stringent
from one sample regression to another, given that the only differences between samples are in the unregulated industries with
which comparisons of regulated prices are being made. Rather,
the rationale might be that, in particular industry regressions,
we are observing price variations having to do with particular
competitive conditions in these samples of unregulated sales
conditions not directly incorporated in measures of elasticity
This rationale takes away credit given to "regulation"
of demand.
that is due to "competition" in the unregulated markets.
—
43
Separation is not possible, because prices are prices,
and the two-part tariffs that are there cannot be eliminated.
Only the crudest of indications can be formulated of the effects
of rate-of-return regulation alone.
from comparing average revenue P
*
This indication follows
the average of initial and
,
commodity charges weighted by volumes delivered, with the un-
regulated unit price P
*
*
.1
,,
;
this price
could be considered
P
the "equivalent" one-part tariff given that the F.P.C. set rates
of return on capital so that the same profits were made as are
now collected under the two-part tariffs.
aP2 + b (ac/'aq
- 'bC/Sq
)
The equation
=
P-,
+ d might show the price-increasing
effects from regulation in the a,b coefficients.
This equation has been fitted to the 137 observation
sample of pooled industrial-retail utility sales in 1969.
The
regression is as follows:
P* = .900P*^ + 1.558(ec/dq^ - «C/2)q
1
1
-
+ 35.585L
)
1
2
65.513L,
2
+ 32.955L_ + 40.261L,
3
4
indicating higher relative prices for regulated sales than shown
in the marginal price comparisons above.
But this is deceiving,
The fitted equation was
P* = l.llOP* - 1.773(E)C/aq
(0.034)
(0.265)
+ 70.723L
(12.830)3
- 36.583L^;
(15.326)^
-•-
ac/^q
)
- 39.509L,
(0.265)
- 39.509L2
(14.572)
44
because the implied values of X have to be much closer to zero
in this case given that the assumed elasticity e
two-part tariff
—
= -3.5,
e
without a
lies in the range from -1.0 to -2.0.
the estimated values of
e
—
X
are positive,
= -2.0 to 7.8 for e
In fact,
ranging from .60 for
= -2.0,
e
= -1.1 so that
the measurable effects from rate regulation alone are perverse.
To be more specific, these effects are contrary to expectations
from the theory of the constrained firm, since there are relative
price reductions on sales to retail ptxblic utilities.
There are of course other alternatives than setting the
calculated
P,
for the pipelines to consider,
the necessity of charging unit prices.
that,
if ever faced with
There is little chance
given a regulatory profit constraint, they would charge
the price exactly equal to the average revenues from the two-
part tariffs.
This price would be too low
—
it now occurs at
elasticities too close to -1.0 to be profit-maximizing
—
so
that there would be more regulatory price increasing effect than
is shown here.
All that can be said with these findings on Pi»P2
is that the effects from profit regulation alone are most likely
negligible, and perhaps even perverse.
45
These views have to be brought together with those in the
*
P,,P2 equations, and on the earlier price equations on industrial
sales free of the effects of regulation.
P
alone are quite tentative,
The assessments of P
,
since it is not possible to define
precisely what is being measured by "price" in the average revenue equations.
however.
They do not contradict earlier assessments,
Overall, the effects of regulation are found in calcul-
ated values of
X
in the range from 0.0 to -0.3 which is rather low
but implies that there has been creation of price disparities by
regulation.
In economic terms, there has been a tendency towards
higher relative prices under regulation, but not a strong tendency.
Prices charged by pipelines on regulated retail utility sales and
on unregulated industrial sales tend to vary with costs of transport, with elasticities of demand, and with the field purchase
prices of gas transported by the different pipelines.
Prices at
the margin on regulated sales are lower when there are two-part
tariffs; but after accounting for the effect of the two-part
tariff, charges for regulated sales are greater than on unregu-
lated sales.
That is, after these strong tendencies for prices
to vary with costs, with demand elasticities and with tariff
structure, prices seem to be only somewhat higher on regulated
46
sales and as expected from the theory of the firm subject to
the
A regulatory constraint.
^J
Date Due
OCT^OJ98Z"
^^<*^^6'85
M^«F
S 79'
WR
^'88
'Ml
Di'mi
SEP 11
icife
Lib-26-67
3
DD3 TDb
TOflO
3
i)
Ii!
!
II
III II
3
I
1
1
TQflO
I
I
II
I
DQ3 a75 3b3
MOfl
HD28
|v
I
I
TDfiD
I
II
I
I
II
I
D03 TDb
I
,
525-7f
3flE
«nuB.....s
M.I.T. Alfred P. Sloan
School of Management
Nos.523-71-«orking Papers.
Nos. 552-71 Nos. 523-71 to Nos. 552-71
.M4li+
iiiiiiiiiii -^3
TOflO
003
fl7S
3T7
527-7/
3
TOflO
003 TOb 3TD
3
TDfiQ
003 ^Ob 3bb
3
TOflO
003
/
BAStMEMf
J-
fi7S
7
413
SIGN
tliis
at
the
card and present
it
with book
CIRCULATION DESK
5!^'7|
3
=1060
MASSACHUSETTS INSTITUTE OF
003 TOb 6Mb
TECHNOLOGY
LIBRAJEQES
cM-71
II
3
TQ60 DQ3
"lOb
ttbJ.
DEWEY LIBRARY
f?^-7'
3
^1060
003 10b 67T
Download